Managing Real Exchange Rate for economic development ... › pdf › v_2017_11_11_missio_1.pdf ·...
Transcript of Managing Real Exchange Rate for economic development ... › pdf › v_2017_11_11_missio_1.pdf ·...
1Planning Department of Minas Gerais (SEPLAG-MG) 2Professor at Federal University of Minas Gerais (CEDEPLAR/UFMG) 3 Professor at Federal University of Minas Gerais (CEDEPLAR/UFMG).
Managing Real Exchange Rate for economic development: empirical
evidences from developing countries
Lúcio Otávio Seixas Barbosa1
Frederico G. Jayme Junior2
Fabrício J. Missio3
Abstract: This paper analyses macroeconomic policies capable of influencing the long-
run real exchange rate (RER). In this vein, it identifies economic policy tools that can
devalue RER, covering a theoretical issue neglected by the economic literature, which
argues that competitive exchange rate enhances growth. After discussing the “Trilemma”,
we identify those variables that could affect RER without constraining monetary policy
or exchange rate regime choice. In what follows, we model the probability of achieving
an undervalued (small or large) RER for a sample of 14 developing countries from 1980
to 2010 (30 years) by applying econometric techniques for discrete choice and censored
data. Afterwards, we compare the results for Latin-American nations with Asian ones.
They suggest that competitive exchange rate requires different approaches depending on
the region. Moreover, Latin American countries need to take on additional policies so that
interventions in the Foreign Exchange Market become effective.
Key words: competitive exchange rate, exchange rate management, exchange rate
policy.
Jel Code: F31; F40; F41
2
1 - Introduction
Recenty many papers have been dealing with economic growth strategies based on
competitive (devaluated) exchange rates1. From a theoretical point of view, there are
different channels through which exchange rate policies enhance economic growth, as
boosting effective aggregate demand and improving functional or inter-sectoral income
distribution (BHADURI e MARGLIN, 1990; ARAUJO, 2012; MISSIO and JAYME JR.,
2012; among others). Empirically, there is robust evidence that competitive exchange rate
prompts economic growth (DOLLAR, 1992; RAZIN and COLLINS, 1997;
ACEMOGLU et al, 2002; GALA 2007b; RODRIK, 2008; MISSIO et al, 2015, among
others).
Nevertheless, though such an approach endorses exchange rate devaluation, it does
not mention how to accomplish it. In other words, few papers disclose economic strategies
to devaluate real exchange rate (RER). Yet, under free capital mobility, is it possible to
adopt exchange rate policies to keep a competitive exchange rate?
In this vein, this paper evaluates the means by which exchange rate policy can affect
RER. It aims at identifying those variables that cause exchange rate devaluation. Indeed,
this is the novelty of it. On one hand, we present an alternative approach so as to identify
such variables. On the other hand, we follow a different empirical approach.
We employ discrete and censured econometric models, which allow assessing
whether exchange rate policies are effective to devaluate RER. Since the dependent
variable is binary (RER is devalued or not) or censured (overvalued periods are reported
as zero), these approach are more appropriate and give better results. Moreover, we test
such models in a panel data context, which are more appealing than Kubota (2011)
approach.
We analyze the results comparing two country groups: Latin-American countries
and Asian ones. By doing so, we seek to understand the relative success at managing the
exchange rate of the latter group, while the former failed. In sequence, we present those
measures that would allow Latin-American countries to take a similar route to Asian ones.
The paper has three sections, besides this introduction and the concluding remarks.
Section 2 discusses exchange rate policies from a theoretical perspective. Section 3
presents how to measure them. Section 4, besides providing the empirical methodology,
shows the analysis of the results.
2 – Exchange Rate Policies: theoretical approach
According to the policy “Trilemma” framework, the monetary authority can
achieve only two out of three policies goals, which are (1) financial integration, (2)
exchange rate stability and (3) monetary autonomy. Exchange rate stability improves
predictability of external trade agreements and lessens foreign direct investment risks;
free capital mobility would improve global savings allocation2; and monetary policy
autonomy allows Central Banks (CBs) to focus on price stability and economic growth.
1 Macroeconomic fundamentals drive the real exchange rate under a conventional approach (see Isard,
2007). Following it, there is no role for exchange rate policy, once it does not affect the real economy. In
contrast, under a non-conventional approach short-run and nominal variables, as well as long run and real
variables, influence the real exchange rate (see Kaltenbrunner, 2015). 2 Kregel (2004) and Palma (2012) argue that growth strategies based on external finance are harmful for
developing countries, since they increase financial fragility (Minskyan approach) and exposes countries to
massive liquidity of financial markets (according to Kindleberger).
3
Thus, when policymakers choose to keep exchange rate stability and free capital
mobility, they step down CB autonomy. Assuming that government bonds issued by
countries are perfect substitutes, a fixed exchange rate regime demands that both rates are
equals, in line with Uncovered Interest Parity (UIP).
Although the policy “Trilemma” framework may be useful for small open
economies, its results must be assessed cautiously3. In fact, the great majority of countries
are not restrained by binary choices. They are able to choose the level of financial
integration with standing capital controls, and to manage their exchange rate under a
manage float regime, while keeping a monetary policy relatively independent. Therefore,
countries seek for ways of matching these policies (Bresser-Pereira, 2012).
In addition, such a framework relies on UIP validity, which has questionable
empirical results (Isard, 2007), and on the controversial assumption of desirable free
capital mobility, which exposes developing countries (DCs) to sudden-stops of capital
flows capable of undermining domestic economy. Hence, countries usually do not work
in the three points of the triangle but somewhere inside it.
Lately, we have seen financial markets becoming even more integrated
internationally, which forces countries to choose down giving up either exchange rate
stability or monetary independence. This does not mean that they cannot have both
simultaneously. Theoretically, there is nothing that prevents a country from pursuing a
manage float exchange rate regime in which half of every fluctuation in demand for its
currency is accommodated by intervention and the other half can be reflected in the
exchange rate4 (Frankel, 1999).
Nevertheless, CB cannot always intervene in the foreign exchange market. For
instance, under an uprising demand for foreign exchange rate, it relies on its currencies
reserves. Furthermore, CB intervention results usually are far more complex than theory
prescription. Taking this into consideration, capital flows depend upon exchange rate
expectations, which in its turn can be changed by the signaling and behavior of CB itself.
According to Frenkel and Taylor (2006, p.8) “The fact that no single form of
transaction or arbitrage operation determines the exchange rate means that monetary
authorities have some leeway in setting both the scaling factor between their country’s
price system and the rest of the world’s, and the rules by which it changes”. This is the
key issue: which are those economic policy instruments not mentioned in the
“Trilemma”5 are capable of influencing RER?6 In what follows, we identify them for
DCs, taking into account theoretical and historical matters.
We rank three economic policy instruments of which CB can enforce only two.
The other one requires CB coordination with government institutions responsible for
economic policy. Therefore, we assigned it as macroeconomic policy coordination.
i) Central Bank (CB) intervention: Since 1990, when capital account was deregulated,
Development Countires (DCs) have increased their international reserves holding. Such
3 Following Frenkel and Taylor (2006), the “Trilemma” does not hold. Even under free capital mobility,
the monetary authority can buy and sell bonds in domestic and foreign markets, or even embrace monetary
control policies, which regulate money supply regardless other forces that drive the exchange rate. 4 First, CB buy dollars, selling domestic currency. Then, the monetary base and international reserves rise.
At the same time, CB sell domestic bonds, returning the monetary base and the interest rate to the initial
value. 5 We discuss instruments beyond the “Trilemma” framework, apart from capital control, once it is not
necessarily desirable. Specifically, we do not discuss exchange rate regime, interest rate level and financial
integration, assuming that other policies allows the combining of these three ones. 6 A key message of the “Trilemma” is the scarcity of policy instruments (Aizenman, 2010).
4
reserves provide self-insurance against volatile finance flows, which are associated with
the growing exposure to sudden-stops and deleveraging crises (Aizenman, 2010).
Aizenman (2010) states that the link between hoarding reserves and financial
integration adds a fourth dimension to the “Trilemma”. In the short-run, hoarding and
managing international reserves increase DCs’ financial stability and capacity to run
independent monetary policy. Latin American countries liberalized their financial
markets rapidly since the 1990s, reducing the extent of monetary independence and
maintaining a lower level of exchange rate stability, whereas Emerging Asian economies
stand out by achieving significant levels of exchange rate stability and growing financial
openness, without giving up greater monetary independence. According to the author, the
main difference between these two groups of economies is the higher level of
international reserves holding of the latter.
Rajan (2012) argues that the higher level of international reserves holding of
Emerging Asian economies suggests that CBs are more sensitive to exchange rate
appreciation (“fear of appreciation”). To avoid it, CBs intervene to defend the exchange
rate, while maintaining monetary autonomy.
Theorethically, the author assumes that CB has control over international reserves
holding. Thus, the monetary authority objective function has two core goals: keeping
optimal international reserves level and exchange rate in its target. If CB intervenes more
when the exchange rate appreciates, it signals its preference over a devalued currency
(Rajan, 2012).
In this context, the central point is to assess CB capability of sterilizing monetary
effects of international reserves holding. For Frenkel (2007), the condition for combining
exchange rate management and monetary autonomy is the existence of an excess supply
of international currency at the exchange rate target. That is, when international liquidity
is high, if CB does not intervene, local currency will appreciate due to favorable
conditions in current and capital accounts. To cope with it, the monetary authority can
achieve the exchange rate target by purchasing the excess of foreign currency supply and
can control the interest rate by issuing treasure or CB’s bonds to sterilize the monetary
effects of this intervention.
However, the author argues that the “Trilemma” holds in situations of deficits,
since the CB ability to intervene in the money market relies on its available amount of
reserves. Therefore, CBs could not adopt the exchange rate target without affecting the
interest rate in situations of excess demand for international currency.
In addition, sterilizing buying operations cannot be applied in every circumstance.
Its sustainability depends on the interest rate earned by the international reserves, on the
local interest rate, on the exchange rate trend and on the evolution of the variables
determining supply and demand of monetary base. Frenkel (2007) concludes that there is
a maximum local interest rate above which the policy of sterilization becomes
unsustainable7.
Broadly, CBs can manage exchange rate by sterilized interventions. Nonetheless,
deficit conditions or sterilization costs bound its action.
.
ii) Capital control: Besides sterilized interventions, capital control measures can help to
manage RER. In fact, it is possible to mix regulated capital markets, which would allow
7 The Brazilian experience between 1994 and 1998 illustrates this issue. The difference between the interest
rate paid by the public debt and the income earned from international reserves discloses the high costs of
sterilization. In addition, since the growth rate of public revenue was much smaller than the interest rate
paid by the public debt, sterilization policy sustainability was undermined (See Palma, 2012).
5
capital flows, though discouraging speculative ones, with exchange rate stability and
monetary autonomy.
Financial flows intensify boom-bust cycles, particular in DCs, as happened in
many Latin American countries, mainly after capital account deregulation (Ocampo,
2003; Palma, 2012). During upswings, the external lending boom enhances domestic
credit and private sector spending. The rise of domestic asset prices, as well as the excess
of liquidity, reinforces credit and spending growth. This behavior also increases the
vulnerability of the financial system during the subsequent downswing, when debtors
have difficulties serving their obligation and it becomes clear that loans did not have
adequate backing.
According to Ocampo (2003), monetary policies have limited degrees of freedom
to smooth out the dynamics of boom-bust cycles irrespective of the exchange rate regime.
Furthermore, the accumulation of foreign currency and maturity mismatches on the
balance sheets of both financial and non-financial agents during upswings are an
additional source of vulnerability. Hence, capital-account regulations potentially have a
dual role: i) as a “liability policy” to improve private sector external debt profiles; ii) and
as a macroeconomic policy that provides some room for counter-cyclical monetary
policies.
The former recognizes that a maturity profile that leans towards long-term
obligations will reduce domestic liquidity risks. In other words, an essential component
of economic policy during booms should be measures to improve the maturity structures
of both the private and public sectors’ external and domestic liabilities. On the equity
side, foreign direct investment should be preferred to portfolio flows, which are more
volatile (Ocampo, 2003).The latter allows monetary authorities to cope with boom-bust
cycles, providing some room to “lean against the wind” during periods of financial
euphoria by adopting contractionary policies and/or reducing exchange rate appreciation
pressures. During crises, capital account regulations improve authorities’ ability to mix
additional degrees of monetary independence with a more active exchange rate policy
(Ocampo, 2003).
In this vein, Magud, Reinhart and Rogoff (2011) claim that capital controls can
prevent exchange rate appreciation. According to them (p.26-27, 2006), “capital controls
on inflows seem to make monetary policy more independent, alter the composition of
capital flow, reduce real exchange rate (although the evidence is more controversial)”,
but “seem not to reduce the volume of net flows (and hence, the current account balance)”.
At the same time, “limiting private external borrowing in the ‘good times’ plays an
important prudential role because more of than not countries are ‘debt intolerant’”.
Therefore, besides preventing exchange rate overvaluation, they contribute to a better
public sector’ debt profile.
Thus, capital account regulations measures are favorable to domestic economy,
though they may not influence RER.
iii) Macroeconomic policy coordination: it connects the fiscal stance with wage
management policies. Assuming that economic policies aim at keeping a competitive
RER, the dispute between capitalists and workers must be considered. Bhaduri and
Marglin (1990) (and its subsequent developments) presented a model for handling such
an issue. In simple terms, a RER devaluation decreases real wage (exchange rate pass-
through), harming the working class, but simultaneously increasing capitalist’s
profitability. On one hand, the fall of the wage mass has a negative impact on aggregate
demand. On the other hand, if the economy is profit led, the most important is to keep
capitalists’ profitability.
6
Theoretically, Rapetti (2011) suggests that exchange rate devaluation diminishes
wages in the tradable sector, prompting output and employment growth (considering a
profit led regime) in both tradable and non-tradable sector. In the medium run, the
profitability rise in the tradable sector would stimulate investments therein (with
increasing returns to scale through learning by doing, for instance), boosting the capital
accumulation process and reinforcing employment, output and aggregate demand growth.
As a result, workers would bargain better wages and non-tradable prices would also grow
fast (income effect), which would hamper tradable sector profitability. Therefore, the
management of productivity gains, labor costs (wages) and non-tradable prices are
fundamental to keep a virtuous cycle.
In this context, the author argues that countries, which intend to implement a
strategy based on competitive RER, need to coordinate macroeconomic policy (exchange
rate and wage management, and monetary and fiscal policy) to avoid a raise in non-
tradable prices and wages. Since in a competitive RER regime monetary policy is
subordinated to exchange rate policy, fiscal policy has a greater responsibility in the
managing of the pace of domestic demand, which can be curbed by rising fiscal surplus.
Instead of it, government may choose to manage real wage, so that labor costs do not rise
above productivity gains of the tradable sector. In sum, a competitive RER policy relies
on macroeconomic policy coordination (Rapetti, 2011).
As ranked above, there are at least three different ways of managing RER through
policy instruments beyond the “Trilemma” framework. Empirically, Kubota (2011) ran
some tests based on the following active economic policies: exchange rate regimes,
capital controls, foreign exchange market intervention, fiscal and external policies
(financial and trade openness), liability dollarization and fiscal surplus. Though the
results are mixed, they point out that foreign exchange rate intervention and fiscal
discipline are effective on small to medium devaluations, i.e., below 20%. Flexible
exchange rate arrangements have a positive impact on pursuing a competitive RER,
irrespective of the RER target, i.e., even for devaluations over 20%. Lastly, portfolio
flows have an adverse effect on devaluations.
Based on Kubota (2011) results, it seems that embracing competitive RER
strategies is not a trivial matter, although Gala (2007a, 2007b) points out that the
outstanding economic growth of Asian economies compared to Latin American ones
were due to exchange rate management. Therefore, disclosing how the former countries
accomplish it is a key issue, particularly considering its contractionary and inflationary
effects and the obstacles to coordinate macroeconomic policy.
3. Empirical indicators and identification strategy
3.1 Empirical indicators
As aforementioned, there at least three ways to prompt competitive RER policies,
circumventing the “Trilemma" approach: intervention in exchange rate markets through
(often-sterilized) reserve accumulation, capital controls’ measures, and wages and
demand management policies. In what follows, we show how to measure each of them:
7
a) Intervention in exchange rate markets: Following Levy-Yeyati et al (2013), we
subtract government deposits at the CB from its net foreign assets8. More specifically, we
define net reserves in dollar in month 𝑚 as:
𝑅𝑚 =
𝐹𝑜𝑟𝑒𝑖𝑔𝑛 𝐴𝑠𝑠𝑒𝑡𝑠𝑚 − 𝐹𝑜𝑟𝑒𝑖𝑔𝑛 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠𝑚 − 𝐺𝑜𝑣 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠𝑚
𝑒𝑚
(1)
where 𝑒 is the dollar price in terms of domestic currency.
Then, we compute the ratio between CB reserves and M2 (M1 plus quasi-money).
𝑅2𝑚 =
𝑅𝑚
𝑀2𝑚
(2)
The intervention measure to country 𝑐 in year 𝑦 is the annual average change of
this ratio.
𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒𝑐,𝑦 = ∑
1
12(𝑅2𝑐,𝑦,𝑚 − 𝑅2𝑐,𝑦,𝑚−1)
12
1
(3)
Thus, a positive integer 𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒 implies a strong degree of intervention, since
the index will be positive only if reserve accumulation exceed the increase of monetary
aggregates. In such cases, CB would by foreign currency to avoid exchange rate overvaluation
(“fear of appreciation”). In other words, CB would buy foreign currency to enlarge its own
reserves and not to widen the monetary base in face of a raise of the demand for money. On the
other hand, negative values of this index indicate that CB is selling reserves to defend domestic
currency from devaluations.
Besides this index, the authors suggest another way to measure intervention in the
exchange rate market:
𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒2𝑐,𝑦 = ∑
1
12
12
1
𝑅𝑐,𝑦,𝑚 − 𝑅𝑐,𝑦,𝑚−1
(𝑀𝐵𝑎𝑠𝑒𝑐,𝑦,𝑚−1
𝑒𝑐,𝑡,𝑚−1)
(4)
in which changes in net foreign reserves are normalized by monetary base. While reserves
accumulation can be explained by exchange rate policy, GDP growth can foster the
demand for money, which in its turn would prompt a rise in precautionary demand for
international reserves. This behavior is linked to crises proximity, currency mismatch and
fear of International Monetary Fund (IMF) – avoid IMF rescuing in cases of foreign
currency shortage. In addition, financial depth (measure by the ratio M2/GDP) would
bring about international reserves accumulation as well, assuming that foreign currency
liquidity works as a bumper in response to the risk of capital flight. In such cases,
𝑖𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒2 would be a biased index due to endogeneity problems, since it does not
differentiate between a raise in international reserves caused by GDP growth and those
associated with interventions in exchange rate9.
b) Capital controls: capital control data are usually assigned by a binary variable from
IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions
(AREAER). Its value is equal to unity when there are restrictions on capital account
8 According to the authors, oil producing countries and countries with important privatization programs are
examples of cases where the latter correction matters. 9 We use this alternative index to verify the effect of intervention in the exchange rate market. The first
index 𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒 is more careful insofar it considers only international reserves accumulation above
monetary aggregate growth.
8
transactions, and zero otherwise. Kubota (2011) claims that this measure does not
consider capital controls intensity.
To circumvent this drawback, we employ Chinn-Ito (2006) index of capital
account openness, which considers different kinds of financial restrictions. Their measure
relies on a dummy variable, which codifies records of restriction on financial transactions
between countries in line with the AREAER. The higher the index, the less the extent of
capital control in that country.
c) Macroeconomic policy coordination: we measure it through two indicators: fiscal
policy and wage management.
The former we measure by government effort to curb its consumption, assuming
that the great majority of government expenditures concern non-tradable goods. Doing
so, it helps to smooth non-tradable inflationary pressures and, consequently, alleviates
RER overvaluation forces. We employ the ratio government final consumption/GDP to
portray it.
The best proxy for wage management would be unit labor costs, which is the
average cost per unit of output, i.e., the relation between wages and productivity.
Nonetheless, this measure is not available for DCs. Therefore, on the one hand, we verify
the labor costs trend – GDP per person employed, and on the other hand, we consider the
unemployment level. When the unemployment rate is high, wages will not be pushed up
as more people are employed.
Thus, from a theoretical and empirical approach, it seems reasonable to infer that
employment growth, in a low labor supply scenario, connected with productivity
slowdown, would hamper a competitive RER strategy. In contrast, if unemployment rates
keep high, there not would be such effects.
3.2 Identification Strategy
The estimation strategy involves a sample of 14 countries, from which 7 are Asian
and the others 7 Latin Americans, based on data from 1980 to 20101011. The goal is to
assess whether macroeconomic policies are able to affect RER, keeping it at competitive
level.
The dependent variable is the extent of RER undervaluation when it takes place
otherwise zero when RER is in equilibrium or overvalued, in line with Rodrik (2008)
index. According to it, the equilibrium RER is computed by amending it to Balassa-
Samuelson (BS) effect, i.e., non-tradable prices are cheaper in DCs.
First, we regress RER12 against GDP per capita (rgdpch)13:
ln 𝑅𝐸𝑅𝑖𝑡 = α +β ln ( 𝑟𝑔𝑑𝑝𝑐ℎ𝑖𝑡) + 𝑓𝑡 + 𝑢𝑖𝑡 (5)
where 𝑖 is an index for countries and 𝑡 is an index for time period, 𝑓𝑡 is the time specific
effect and 𝑢𝑖𝑡 is the idiosyncratic error. This regression yields an estimated �̂� = 0.18 with
a very high absolute value of t-statistic around 42.
10 Asian countries: China (CHN), Hong Kong (HKG), South Korea (KOR), Indonesia (IDN), Malaysia
(MYS), Singapore (SGP). Latin American countries: Argentina (ARG), Brazil (BRA), Chile (CHL),
Mexico (MEX), Peru (PER), Venezuela (VEN) and Colombia (COL). 11 Some data (particularly, exchange rate regime) are available only until 2010. 12 The proxy variable to RER is the inverse of pl_gdpo (Price Level of Output-side real GDP at current
PPPs in mil. 2005US$) from Penn World Table 8.1 13.This variable is from Penn World Table 8.1 (rgdpo/pop).
9
The next step is to calculate the difference between the actual RER and the BS
adjusted rate:
ln 𝑈𝑁𝐷𝐸𝑅𝑉𝐴𝐿𝑖𝑡 = ln 𝑅𝐸𝑅𝑖𝑡 − 𝑙𝑛𝑅𝐸𝑅𝑖𝑡̅̅ ̅̅ ̅̅ ̅̅ (6)
where 𝑙𝑛𝑅𝐸𝑅𝑖𝑡̅̅ ̅̅ ̅̅ ̅̅ is the predicted values from equation (5). Whenever ln 𝑈𝑁𝐷𝐸𝑅𝑉𝐴𝐿14 is
positive, it indicates that the exchange rate is set so that goods produced at home are cheap
in dollar terms: the currency is undervalued. When ln 𝑈𝑁𝐷𝐸𝑅𝑉𝐴𝐿 is below unity, the
currency is overvalued.
The explanatory variables are the economic policy instruments used to promote
RER undervaluation. Besides intervention in exchange rates (interv.exrate), capital
controls (cap.control), government expenditures (spend.gov), worker productivity
(produtiv) and unemployment rate (unemp.rate), we use control variables representing
those in the Trilemma framework: exchange rate regime (dummy.fixed and
dummy.intermed), financial integration (finan.integ) and interest rate (i). These control
variables can affect RER due to the following reasons: i) RER and interest rate are linked
(Barbosa, Jayme Jr and Missio, 2017); ii) and exchange rate regime (Libman, 2017) and
financial integration influences the ability to keep a competitive RER (Kubota, 2011).
The general form of the equation to be estimated is:
ln 𝑈𝑛𝑑𝑒𝑟𝑣𝑎𝑙𝑖,𝑡 = 𝛽0 + 𝛽1𝑖𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒𝑖,𝑡 + 𝛽2𝑐𝑎𝑝. 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡
+ 𝛽3𝑠𝑝𝑒𝑛𝑑. 𝑔𝑜𝑣𝑖,𝑡 + 𝛽4𝑝𝑟𝑜𝑑𝑢𝑡𝑖𝑣𝑖,𝑡 + 𝛽5𝑢𝑛𝑒𝑚𝑝. 𝑟𝑎𝑡𝑒,𝑡
+ 𝛽6𝑖 + 𝛽7𝑓𝑖𝑛𝑎𝑛. 𝑖𝑛𝑡𝑒𝑔𝑖,𝑡 + 𝛽8𝑑𝑢𝑚𝑚𝑦. 𝑓𝑖𝑥𝑒𝑑𝑖,𝑡
+ 𝛽9𝑑𝑢𝑚𝑚𝑦. 𝑚𝑎𝑛𝑎𝑔.𝑖,𝑡 +𝜇𝑡+𝜂𝑖+𝜀𝑖,𝑡
(7)
Table 1 summarizes the list of variables.
Table 1: List of the variables in the research
Abbreviaton Variable Comments Unit of
measurament Source
underval Real exchange rate
undervaluation
Undervaluation index following
Rodrik (2008) Index
Own elaboration
based on PWT 8.1
underval1 Real exchange rate
undervaluation
Undervaluation index following
MacDonald and Vieira (2012) Index
Own elaboration
based on PWT 8.1
interv.exrate Intervention in
exchange rate
Index of variation of the ratio
international reserves/M2 Index
Own elaboration
based on IFS data
interv.exrate2 Intervention in
exchange rate
Index of variation of international
reserves normalized by monetary base Índex
Own elaboration
based on IFS data
cap.control Capital controls Financial openness index Index Chinn-Ito (2006)
spend.gov Government
expenditures
Government final consumption
expenditures normalized by GDP % IFS
Produtiv GDP per worker GDP per worker (productivity proxy –
rgdpo/emp)
US$ Dollar
2005 PWT 8.1
unemp.rate Unemployment rate Unemployed rate (% of work force) –
national estimates % WDI
i Interest rate Interest rate set by monetary authority % IFS
14 Underval is in logarithmic form. For robustness check purposes, we will employ Macdonald and Viera’s
(2012) index of undervaluation.
10
Abbreviaton Variable Comments Unit of
measurament Source
finan.integ Financial integration
External liability normalized by GDP.
It includes the stock of debt and
portfolio equity liabilities, foreign
direct investment, debts and financial
derivatives
%
Own elaboration
based on Lane e
Milesi-Ferreti
(2007)
dummy.fixed Fixed exchange rate
regime
Classification of exchange rate regime
in line with Ilzetzki,Reinhart and
Rogoff (2008) methodology and
grouped following Coudert and
Couharde (2009)
Discrete
variable
Ilzetzki,Reinhart e
Rogoff (2008)
dummy.intermed.
Intermediate
exchange rate regime
Intermediário
Classification of exchange rate regime
in line with Ilzetzki,Reinhart and
Rogoff (2008) methodology and
grouped following Coudert and
Couharde (2009)
Discrete
variable
Ilzetzki,Reinhart e
Rogoff (2008)
Note: IFS – International Financial Statistics; WDI – World Development Indicators; PWT – Penn World Table
Source: Own elaboration.
To test our model we employ discrete choice models (logit) and censored
regressions15 (tobit) within an unbalanced panel dataset. In the former case, the potential
outcome is discrete. In the latter case, we do not observe values of the dependent variable
above or below certain magnitude owing to a censoring or truncation mechanism.
In discrete choice models, underval takes on the value 1 or 0 depending on the
threshold set. We set three thresholds: RER is undervalued irrespective to its extent, RER
is over 10% undervalued (modest undervaluation), or RER is over 20% undervalued
(significantly undervaluation). In censored regression, whenever RER is overvalued, it is
recorded as zero.
In panel data sets, the presence of individual effects complicates matters
significantly in face of the incidental parameter problem. Application of these models
become more complex due to non-observed heterogeneity (individual fixed effects)16.
Logit Model
For logit models, Chamberlain (1980) suggests maximizing the conditional
likelihood function to obtain the conditional estimates for 𝛽. In order to test for fixed
individual effects one can perform a Hausman-type test based on the difference between
Chamberlain’s conditional maximum likelihood estimation (MLE) and the usual logit
MLE. The latter estimation is consistent and efficient only under the null hypothesis of
no individual effects. Chamberlain’s estimator is consistent whether 𝐻0 is true or not, but
it is inefficient under 𝐻0 because it may not use all data (Baltagi, 2005).
It is important to highlight that conditional logistic regression differs from
ordinary logistic regressions in that the data are divided into groups and, within each
group, the observed probability of a positive outcome is either predetermined due to data
construction or in part determined because of unobserved differences across the groups.
15 Censored regression can be put into one of two categories. In the first case there is a variable with
quantitative meaning, 𝑦∗, and we are interested in the population regression, 𝐸(𝑦∗ ∥ 𝑥). However, 𝑦∗ is not
observable for part of the population. A second kind of application of censored regressions is where the
dependent variable, 𝑦, is an observable outcome, which takes on the value zero with positive probability
but it is a continuous random variable over strictly positive values. This is an example of corner solution
outcome. In such cases, we are interested in features of the distribution of 𝑦 given 𝑥, sucha as 𝐸(𝑦 ∥ 𝑥) and
𝑃(𝑦 = 0 ∥ 𝑥) (Wooldridge, 2002). 16 See Wooldridge (2002), Greene (2003) and Baltagi (2005).
11
Thus, the likelihood of the data depends on the conditional probabilities, i.e., the
probability of the observed pattern of positive and negative responses within a group is
conditional on the number of positives outcomes being observed. Terms that have a
constant within-group effect on the unconditional probabilities – such as intercepts and
variables do not vary – cancel in the formation of these probabilities and so remain not
estimated.
Groups that contain all-positive or all-negative outcomes provide no information
because the conditional probability of observing such groups is 1 regardless of the values
of parameters 𝛽. Thus, they are dropped off estimations.
Therefore, first, we run the Hausman test to decide which model is more
appropriate. Then, we run the estimations, bearing in mind that, in logistic models, the
exponentiated coefficients are interpreted as odds ratio, i.e., 𝑜𝑑𝑑𝑠 (𝑖𝑓 𝑡ℎ𝑒 𝑐𝑜𝑟𝑟𝑒𝑠𝑝𝑜𝑛𝑑𝑖𝑔𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑖𝑠 𝑖𝑛𝑐𝑟𝑒𝑚𝑒𝑛𝑡𝑒𝑑 𝑏𝑦 1)
𝑜𝑑𝑑𝑠 (𝑖𝑓 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑛𝑜𝑡 𝑖𝑛𝑐𝑟𝑒𝑚𝑒𝑛𝑡𝑒𝑑) , or, equivalently:
𝑝(𝑒𝑣𝑒𝑛𝑡 ∥ 𝑥 + 1)1 − 𝑝(𝑒𝑣𝑒𝑛𝑡 ∥ 𝑥 + 1)
)
𝑝(𝑒𝑣𝑒𝑛𝑡 ∥ 𝑥)1 − 𝑝(𝑒𝑣𝑒𝑛𝑡 ∥ 𝑥)
) (8)
where 𝑝 is the probabity of event, and 𝑝/(1 − 𝑝) is the called the odds of an event.
Hence, if the coefficient is 1.5, then the odds of the event are 50 percent greater
when the economic policy is on. If it is 0.5, then the odds halves as 𝑥 increase.
Tobit model
In tobit models the individual fixed effect cannot be swept away and as a result 𝛽
and 𝜎𝜐2 cannot be consistently estimated for 𝑇 fixed, since the inconsistency is transmitted
to 𝛽 and 𝜎𝜐2.In order to overcome this issue, Honoré (1992) suggest trimmed least square
estimators. These are semiparametric estimators with no distributional assumptions on
the error term.
Since it removes the fixed effect, it affects marginal effects calculation.
Nonetheless, the author suggests that marginal effects can be computed from the observed
coefficients multiplied by the number of non-censured observations, i.e., 𝛽𝑗 ∗
𝑃𝑟𝑜𝑏(𝑦 > 0). Hence, keeping other variables constant, we have the marginal effect of
the 𝑗 − 𝑡ℎ element of 𝑥.
Thereby, the econometric tests will involve conditional logit estimators and
Honoré’s (1992) approach. For robustness check, we will test the same models with
𝑢𝑛𝑑𝑒𝑟𝑣𝑎𝑙 and 𝑢𝑛𝑑𝑒𝑟𝑣𝑎𝑙1, and with 𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒 and 𝐼𝑛𝑡𝑒𝑟𝑣. 𝑒𝑥𝑟𝑎𝑡𝑒2.
4 Results
The first step is to verify RER behavior in Latin American countries compared to
Asian ones. Following Rodrik (2008) procedure, we notice that there were periods in
which it was undervalued, while in others it was overvalued. Particularly, from 1970 to
2009, based on five-year average, the Argentine Peso was overvalued, whereas the
Colombian Peso was undervalued or near equilibrium (Figure 1).
12
Figure 1: Real Exchange Rate undervaluation: selected Latin American countries (1970-
2009)
On the other hand, Asian countries display a RER devaluation trend (Figure 2).
Except for Singapore and South Korea (and, partially, Hong Kong), their currency were
undervalued. Thus, there are strong indications that they keep a competitive RER policy
over this period.
Figure 2: Real Exchange Rate undervaluation: selected Asian countries (1970-2009)
Hence, it is important to figure out which mechanisms these countries used to
manage its RER. As aforementioned, we are interested in three channels: i) intervention
in exchange rate; ii) capital control; iii) fiscal policy and wage management.
4.1. Econometric Results
Concerning the broad sample (all countries), the results show that the smaller the
capital controls (or the bigger the financial openness index), the smaller is the probability
-140%
-120%
-100%
-80%
-60%
-40%
-20%
0%
20%
40%
60%
19
70
-19
74
19
74
-19
79
19
80
-19
84
19
84
-19
89
19
90
-19
94
19
94
-19
99
20
00
-20
04
20
04
-20
09
ARG BRA CHL COL MEX PER VEN
-60%
-40%
-20%
0%
20%
40%
60%
80%
100%
19
70
-19
74
19
74
-19
79
19
80
-19
84
19
84
-19
89
19
90
-19
94
19
94
-19
99
20
00
-20
04
20
04
-20
09
CHN HKG IDN MYS KOR SGP THAI
13
of attaining a competitive RER. Given that Chinn-Ito index ranges from -1.88 to 2.38, the
odds of significant undervaluation halve17 as it increases by 1 – Table 2 (estimative V).
Robustness check tests indicate that interventions in the exchange rate can help to
sustain a competitive RER policy. We must note that the variable, which stands for
intervention in the exchange rate, has mean zero with a standard deviation about 0.005.
Then, the OR value is not intuitive. Therefore, its direction matters most: whenever above
1, CB intervenes to prompt devaluations.
Productivity increase has almost nil effect on the odds of a RER undervaluation.
Though unemployment rate was significant (estimative III), it goes in the opposite
direction, decreasing by 40% the odds of a RER undervaluation. Hence, wage
management policies may not be a necessary condition to attain a competitive RER.
Nevertheless, fiscal policy may help to sustain it, according to estimates (IV) and (VI).
Regarding the control variables, financial integration coefficients were significant
in estimates (III) and (V), enlarging about 10% the OR. In other words, it seems to
contribute to undervaluing RER. These results may be inconsistent, since the financial
openness index has the opposite effect, or alternatively countries may should seek a way
to take advantage of financial integration, but at the same time set strict rules to evade
speculative capital.
Fixed exchange rate regime is associated with smaller odds of RER
undervaluation. So, such regimes may be engendered to avoid inflation through RER
overvaluation. Finally, interest rates were not significant.
Table 2: Determinants of the probability of keeping RER undervaluation above 0%,
10% and 20% - Condition logit and logit estimations
Underval >
0%
Underval1>
0%
Underval >
10%
Underval1>
10%
Underval>
20%
Underval1>
20%
OR
(I)
OR
(II)
OR
(III)
OR
(IV)
OR
(V)
OR
(VI)
interv.exrate 0.0530 - 25355.2312 - 191.2422 -
interv.exrate1 - 421.1430* - 1.5414 - 59.3337
cap.control 0.6239*** 0.5569*** 0.3472*** 0.7778 0.5524** 0.2292***
spend.gov 0.9138 0.9079 0.9407 0.8515* 0.9095 0.7774**
Produtiv 1.0001*** 1.0000 1.0001* 0.9999* 1.0000 0.9995***
unemp.rate 0.9725 1.0238 0.6638*** 0.9807 0.8697 0.9164
i 0.9996 0.9997 0.9993 0.9999 0.9992 0.9707*
finan.integ 1.0013 1.0111 1.0997*** 1.0188 1.0448*** 0.9917
dummy.fixed 0.0200*** 0.0701*** 0.0000 0.0511** 0.1000** 0.7915
dummy.intermed. 0.3688 0.6648 0.9458 0.3703 0.4557 6.1331
Wald ( χ2 ) 63.54*** 36.87*** 123.44*** 26.71*** 75.77*** 174.91***
Hausman ( χ2 ) 31.19*** 20.98*** 10.98** 117.74** 22.22***
N. 269 271 269 266 302 339
Notes: * p<0.10, ** p<0.05, *** p<0.01. i)Wald test: H0: all the coefficients are zero; ii) Teste de Hausman: H0:Difference
in coefficients not systematic ; iii) OR (odds ratio)); iv) time dummies are not significant in any estimates; v) in estimates
(I) e (III) 63 observations were dropped because China, Malaysia and Thailand currencies were more than 10%
undervalued; in estimate (V) 30 observations were dropped because Korea’s currency were not more than 20%
undervalued; vi) estimate (VI) corresponds to standard logit, since due to low variability of the dependent variable, it was
not possible to estimate by conditional logit. .
17 A one-unit increase corresponds near 25% of Chinn-Ito index.
14
Table 3 displays the results for Latin American countries. In general, they are very
similar to the results from the whole sample. We highlight two points: the coefficients of
interventions in the exchange rate and the coefficients of exchange rate dummies are not
significant.
Table 3: Determinants of the probability of keeping RER undervaluation above 0%,
10% and 20% in Latin American countries - Condition logit and logit estimation
Underval>
0%
Underval1>
0%
Underval >
10%
Underval1>
10%
Underval>
20%
Underval1>
20%
OR
(I)
OR
(II)
OR
(III)
OR
(IV)
OR
(V)
OR
(VI)
interv.exrate 494079.6332 2549929.9301 7.7777
interv.exrate1 1.3367 0.4087 7600236.5644
cap.control 0.4676*** 0.4617*** 0.2668*** 0.4825*** 0.3727*** 0.0757
spend.gov 1.0590 0.9373 0.8971 0.7732*** 0.7963** 0.1612**
Produtiv 1.0001** 1.0000 1.0001 0.9999** 0.9999** 0.9998
unemp.rate 0.6400*** 1.0734 0.6424*** 1.2299*** 1.0924 0.7768
I 0.9996 0.9998 0.9992 0.9995 0.9983 0.9497
finan.integ 1.1003*** 1.0060 1.1157*** 0.9983 1.0176* 0.8457*
dummy.fixed 0.2441 - 0.0000 - - -
dummy.intermed. 0.5688 1.2515 2.9161 2.7442 1.1250 1.1323
Wald ( χ2 ) 67.54*** 32.69*** 76.85*** 36.27*** 47.17*** 60.13***
Hausman ( χ2 ) 31.19*** 16.93*** 11.35
N. 175 155 175 155 149 155
Notes: * p<0.10, ** p<0.05, *** p<0.01. i)Wald test: H0: all the coefficients are zero; ii) Teste de Hausman:
H0:Difference in coefficients not systematic ; iii) OR (odds ratio)); iv) time dummies are not significant in any
estimation;; v) estimative (V) corresponds to standard logit, because Hausman H0 is not rejected; vi) estimates (II), (IV)
and (VI) corresponds to standard logit, since due to low variability of the dependent variable, it was not possible to
estimate by conditional logit; and the variable dummy.fixed is not estimated, because it provides devaluation odds
perfectly (if it equals zero, the currency is undervalued); 26 observations were dropped.
Besides, the unemployment rate has an ambiguous effect: according to estimate
(III), as it increases, the odds ratio of RER undervaluation decreases, while the opposite
effect takes place in estimate (IV). Similarly, the financial integration outcomes are
ambiguous, looking at estimates (V) and (VI).
Table 4 shows the results for Asian countries. It should be noted that conditional
logit estimates, apart from estimate (I), do not reach a maximum point. This happens due
to collinearity issues regarding those variables that stand for the real exchange rate regime
(fixed and intermediated) within the same group. Since we cannot reject the null-
hypothesis of the model consistency under a logistic model by Hausman test, we estimate
all of them according to unconditional logistic approach.
15
Table 4: Determinants of the probability of keeping RER undervaluation above 0%,
10% and 20% in Asian countries – Logit estimations
Underval >
0%
Underval1>
0%
Underval >
10%
Underval1>
10%
Underval
>20%
Underval1
> 20%
OR
(I)
OR
(II)
OR
(III)
OR
(IV)
OR
(V)
OR
(VI)
interv.exrate 0.0040 56327.9928 9.2119e+23
interv.exrate1 151831.6944*
**
866.3127 47.2845
cap.control 1.2330 4.2175*** 1.4090 4.2463*** 1.3123 0.2869
spend.gov 1.3662*** 1.6179** 1.4470*** 1.2561 1.2641** 1.4922
Produtiv 0.9999*** 0.9997*** 0.9999*** 0.9996*** 0.9999*** 0.9993***
unemp.rate 1.1122 0.9546 0.9317 0.6670** 0.9418 0.6988**
i 1.0265 0.9335 1.0066 0.8542* 0.9702 1.1055
finan.integ 1.0058** 1.0098*** 1.0070*** 1.0055 1.0082*** 1.0337**
dummy.fixed 0.0000*** 5.9440 0.0000 0.0322 0.8157 0.3557
dummy.intermed. 0.0000*** 0.0443*** 0.0000 0.0105* 0.1571 4.4468
Wald ( χ2 ) 472.46*** 30.83*** 55.73*** 141.26*** 53.41*** 136.98***
Hausman ( χ2 ) 2.72 - -
N 157 158 157 158 157 158
Notes: * p<0.10, ** p<0.05, *** p<0.01. i)Wald test: H0: all the coefficients are zero; ii) Teste de Hausman:
H0:Difference in coefficients not systematic ; iii) OR (odds ratio)); iv) time dummies are not significant in any
estimation;; v) estimative (I) corresponds to standard logit, because Hausman H0 is not rejected vi) the remaining
estimates corresponds to standard logit, because conditional logit is not estimable.
Capital controls are not favorable to competitive RER policies – estimates (II) and
(IV), i.e., financial openness stimulates it. In fact, even financial integration may be
worthwhile. Intervention in exchange rate plays an important role as well – estimative (I).
Government expenditures instead of harming a competitive RER strategy seems
to foster it. The odds ratio rises near 30% for each 1% rise in government expenditures,
depending on the devaluation RER level. Finally, a rise in interest rates and adoption of
an intermediate exchange rate regime have an adverse effect on undervaluation RER
policies.
Comparing the results of both group, there are substantial differences. First,
capital controls and restrictive fiscal policies were important to Latin American countries,
but not to Asian ones. Second, intervention in the exchange rate had the expected effect
only in the latter group. For both groups we cannot be certain that either rise in wages
above productivity or the exchange rate regime chose were critical to the implementation
of competitive RER strategy.
In face of such results, it is demanding to disclose theoretical underlying reasons
that support them. Dooley, Folkerts-Landau and Garber (2003, 2014) suggests that the
Bretton Woods System does not evolve, it just occasionally reloads a periphery. In the
1950s, the United States (US) was the center region with essentially uncontrolled capital
and goods markets. Europe and Japan, whose capital had been destroyed by the war,
constituted the emerging periphery. These countries chose a development strategy of
undervalued currency, controls on capital flows and trade, reserve accumulation, and the
16
use of the center region as a financial intermediary that lent credibility to their own
system. The US, in turn, lent long-term capital to periphery through FDI mostly.
The Asian countries strategy resembles Europe and Japan. They undervalued their
currency, accumulated reserves and encouraged growth through exportation. On the other
hand, Latin American countries chose to close their markets and foster a system of import
substitution. Such strategy was inhospitable to trade and to the importation of long-term
capital flow. Moreover, it spurred a local production of goods that could not compete
globally, building an inefficient capital stock that would end by having little global value.
In the beginning of the 1990s, following the Washington Consensus, these countries
joined the center directly, opening their economies to trade and their capital market to
foreign capital (Dooley, Folkerts-Landau and Garber, 2003).
The authors argue that Asian countries replaced Europe and Japan. Their
framework includes a “trade account region”, a “capital account region” – Europe,
Canada and Latin American – and a “center country”, the US. The main concern of the
trade account region is to export to the center country. To do so, CBs have consistently
intervened to limit appreciation of their currencies. In contrast, capital account regions,
like the Latin American, opted for a financial integration where private investors are
concerned about the risk/return of their international investment position. To escape from
low yields, they invest in the region, appreciating domestic currencies and discouraging
export. Lastly, the US is the intermediary of the system, which wants finance for its own
growth and foreign savings to spur capital formations.
Hence, while in Asia the probability of keeping an undervalued currency is
positively associated with intervention in the exchange rate market and financial
openness, in Latin American countries it is the other way around. In other words, the first
group, which carried on an export-led strategy, benefited from foreign capitals, whereas
the second one, which drew mainly speculative capital, were not able to intervene in
exchange rate market.
Moreover, there is an interest rate ceiling above which sterilized interventions
cannot be applied (Frenkel, 2006). The monetary authorities in Latin American countries,
even not considering hyperinflation periods in Brazil, Argentina and Peru, set their
interest rate well above from the Asian ones. For instance, from 2000 to 2010 the
difference between them was in average 5% per year. On the other hand, the latter group
accumulated much more reverses. The difference is about 30 billion (excluding China),
increasing particularly after capital account liberalizations in the 1990s. Therefore,
intervention in exchanged rate markets in Asian countries relied on low interest rate and
massive international reserves.
Last but not least, it is important to highlight that competitive RER strategy is the
main common feature in Asian countries growth. Besides that, Akiuz et al (1998) argues
that different routes were taken based on state intervention. From imposing restrictions
on luxury goods consumption, subsidizing key sectors, investing in education and
technology, to establishing join ventures, the role played by government was crucial.
Thereby, government expenditures when administrated by a competent bureaucracy,
relatively insulated from political pressures, can even help RER competitive policies.
The second test relies on Tobit estimation following Honore’s approach. We
assume that there are non-observed effects for each country (historical, cultural and
social) correlated with explanatory variables. Table 5 presents the average marginal
effects of the regressions.
17
Table 5: Determinants of the probability of keeping RER undervaluation – Tobit
estimation
Average Marginal effects (Tobit fixed effects)
Broad sample Latin American countries Asian countries
Underval
(I)
Underval1
(II)
Underval
(III)
Underval1
(IV)
Underval
(V)
Underval1
(VI)
interv.exrate -0.1798 0.0115 - 2.3029* -
interv.exrate1 - 0.1481** - 0.1388 - 0.1907**
cap.control -0.0155 0.0044 -0.0413*** -0.0152*** -0.0218* -0.0084
spend.gov -0.0048 -0.0077 -0.0095* -0.0104*** 0.0093 0.0100*
Produtiv 0.0000 0.0000* 0.0000 0.0000 0.0000*** 0.0000
unemp.rate -0.0054 -0.0064*** -0.0066 -0.0033 -0.0040 -0.0046**
i -0.0001* 0.0000 0.0000 0.0000 0.0092*** 0.0063***
finan.integ 0.0005 0.0004 0.0020** 0.0013*** 0.0003*** 0.0003
dummy.fixed -0.1173 -0.0736 -0.1212*** -0.1079 -0.1192** -0.0819
dummy.intermed. -0.0878* -0.0603 -0.0111 -0.0023 -0.0989*** -0.0603*
Wald ( χ2 ) 65.11*** 26.99*** 201578.32*** 595.65*** 988.11*** 496.41***
Censured
Fraction
0.55 0.41 0.45 0.32 0.66 0.51
N. 332 339 175 188 157 158
Notes: * p<0.10, ** p<0.05, *** p<0.01; i) Wald test: H0: all the coefficients are zero.
The results are quite similar. For Latin American countries, the main difference is
that the fixed exchange rate regime has a negative effect – near 12% – on RER
undervaluation. This indicates that such a regime is associated with strategies to
circumvent inflation pressures. Since 1980, “importing” credibility through fiscal and
monetary discipline has been its main advantage.
For Asian countries, the coefficient of financial openness has the opposite signal,
contributing around 2% to lessen RER undervaluation extent. Indeed, this result is more
in line with the theoretical background, which states that capital controls can be a useful
tool to avoid RER overvaluation.
In sum, we endorse the following results: intervention in exchange markets helps
to keep an undervalued RER; though capital controls have mixed results, they are
important under strategies based only on financial integration (“capital account regions”);
wage management policies are not significant; and expansionary fiscal policies are not
necessarily harmful to competitive RER strategies.
4 Concluding Remarks
In this paper, we assess ways of implementing a competitive RER policy. Looking
at Asian countries growth, which relied on undervalued currencies, we seek to identify
those channels that allowed them to manage their RER successfully. In contrast, we
compare them with Latin American countries, which did not embarked on such strategy,
attaining low growth rates.
First, the theoretical background disclose that most countries combines different
levels of financial openness with a dirty float exchange rates regime and a relatively
18
autonomous monetary policy. Through the analyses, three economic policy tools emerge
as shortcuts to keep a competitive RER strategy: intervention in exchange rate markets,
capital controls, wage management and restrictive fiscal policies aiming at maintaining
tradable sector profitability.
The empirical results point out that intervention in the exchange rate in order to
accomplish an undervalued RER policy prove to be an effective tool for Asian countries.
Restrictive capital controls policies does not seem to be a general rule. Countries, which
followed an export-led strategy, were able to keep a competitive RER, even with greater
financial openness – from a greater (Hong Kong) to a lesser extent (China). Hence, they
become an export platform to US and have accumulated massive international reserves
to manage RER.
In contrast, Latin American countries, which only integrated financially, were
more susceptible to speculative capital and RER overvaluation pressures. Moreover, the
lack of enough reserves, associated with high interest rates, hampered competitive RER
policies.
Overall, managing wages is not critical in the maintaining of an undervalued
currency. Fiscal policies, in their turn, can help or harms such a strategy. In fact,
restrictive fiscal policies seems to be important in Latin American countries, while the
opposite applies in Asian ones. A possible explanation is the role played by each
government in promoting growth.
In sum, if Latin American countries want to follow a similar route to Asian ones,
they need to impose restrictive capital controls and intervene in exchange rate markets by
increasing their international reserves and lowering interest rates. Also, they might should
keep government consumption under control.
References
ACEMOGLU, D.; SIMON, J.; JAMES, R. 2002. Institutional causes, macroeconomic
symptoms: volatility, crisis and growth. Cambridge MA: MIT.
AINZEMAN, J. 2010. The impossible trinity (aka the policy trilemma). Working Paper
Series, UC Santa Cruz.
ARAUJO, R. A. 2012. Efeitos de variações cambiais em uma abordagem de dinâmica de
mudança estrutural à teoria do crescimento restrito pelo balanço de pagamentos.
Economia e Sociedade, 21(3): 585-606.
BALTAGI, H. B. 2005. Econometric Analysis of Panel Data. 3rd ed. John Wiley & Sons.
BARBOSA, L. O. S; JAYME JR., F. G.; MISSO, F. 2017. The determinants of real
exchange rate: a theoretical and empirical approach, International Review of Applied
Economics (forthcoming).
BHADURI, A.; MARGLIN, S.1990. Unemployment and the real wages: the economic
basis for contesting political ideologies. Cambridge Journal of Economics, 14(4):375-93.
BRESSER-PEREIRA, L. C. 2012. For a heterodox mainstream economics: an academic
manifesto. Journal of Post Keynesian Economics, 35(1): 3-20.
CHAMBERLAIN, G. 1980. Analysis of covariance with qualitative data. Review of
Economic Studies, 47(1):225-38.
CHINN, M. D.; ITO, H. 2006. What Matters for Financial Development? Capital
Controls, Institutions, and Interactions, Journal of Development Economics, 81 (1):163-
92.
COUDERT, V.; COUHARDE, C. 2009. Currency Misalignments and Exchange Rate
Regimes in Emerging and Developing Countries, Review of International Economics,
17(1):121-36.
19
DOLLAR, D. 1992. Outward-oriented developing economies really do grow more
rapidly: evidence from 95 LDCs, 1976–1985, Economic Development and Cultural
Change, 40(3):523–44.
DOOLEY, M.P., LANDAU-FOLKERTS, D., GABER, P. 2003. An essay on the revived
Bretton Woods System, NBER, 11p. (Working paper, 9971).
————. 2014. The Revived Bretton Wood Systems’ First Decade, NBER, 40p.
(Working paper, 20454).
FRANKEL, J. A. 1999. No single currency is right for all countries or at all times. NBER
Working Paper Series 7338.
FRENKEL, R.2007. The sustainability of monetary sterilization policies. Cepal Review,
93:29-36.
FRENKEL, R.; TAYLOR, L. 2006. Real Exchange Rate, Monetary Policy and
Employment, DESA Working Paper No. 19.
GALA, P. 2007a. Dois padrões de Política Cambial na América Latina e no Sudeste
Asiático. Economia e Sociedade, 16(1):65-91.
————. 2007b. Real Exchange Rate Levels and Economic Development: theoretical
analysis and econometric evidence. Cambridge Journal of Economics, 32(2):273-288..
GREENE H., W. 2003. Econometric Analysis, 5a Ed. Prentice Hall.
HONORÉ, BO. E. 1992. Trimmed lad and least square estimation of truncated and
censored regressions models with fixed effects, Econometrica, 60(3):533-65.
————. 2008. On marginal effects in semiparametric censored regressions models,
available in htttp://ssrn.com/abstract=1394384.
ILZETZKI, E.; REINHART, C.; ROGOFF, K. 2008. Exchange Rate Arrangements Into
The 21St Century: Will The Anchor Currency Hold?, Unpublished.
Isard, P. 2007. Equilibrium Exchange Rates: Assessment Methodologies, IMF working
paper, WP/07/296.
KREGEL, J. A. 2004. Using Minsky to analyze the impact of international development
finance on international financial stability. Working Paper No. 33.
KUBOTA, M. 2011. Assessing real exchange rate misalignments. Policy Research
Working Paper, 5925.
LANE, P.R.; MILESI-FERRETI, G. M. 2007. The external wealth of nations mark II:
Revised and extended estimates of foreign assets and liabilities, 1970–2004. Journal of
International Economics 73(2):223-50.
KALTENBRUNNER, A. 2015. A post Keynesian framework of exchange rate
determination: a Minskyan approach, Journal of Post Keynesian Economics, 38(3):426-
48.
LEVY-YEYATI, E.; STURZENEGGER, F.; GLUZMANN, P. A. 2013. Fear of
Appreciation. Journal of Development Economics, 101:233-247.
LIBMAN, E. 2017.The effect of exchange rate regimes on Real Exchange Rate
Misalignment, International Review of Applied Economics (forthcoming).
MACDONALD, R.; VIEIRA, F.V.2012. A panel data investigation of real exchange rate
misalignment and growth, Estudos Econômicos, 42(3):433-56.
MAGUD, N.E.; REINHART, C.M.; ROGOFF, K.S. 2011. Capital controls: myth and
reality – a portfolio balance approach. Nber Working Paper Series 16805.
MISSIO, F.; JAYME JR., F. G. 2012. Structural Heterogeneity and Endogeneity of
Elasticities on the Balance of Payments Constrained Growth Model. In: Souziakis, Elias;
Cerqueira, Pedro. (Org.). Models of Balance of Payments Constrained
Growth.1ed.London: Palgrave, v. 1, p. 239-267.
MISSIO, F. J.; JAYME JR., F. G.; BRITTO, G.; OREIRO, J. L. 2015. Real Exchange
Rate and Economic Growth: New Empirical Evidence. Metroeconomica, 66(4):686-714.
20
OCAMPO, J. A. 2003. Capital account and countercyclical prudential regulations in
developing countries, From Capital Surges to Drought, Palgrave Macmillan UK, p.217-
244
PALMA, J. G. 2012. How the full opening of the capital account to highly liquid financial
markets led Latin American to two and a half cycles of ‘mania, panic and crash’.
Cambridge Working Papers in Economics 1201.
RAJAN, R.S. 2012. Management of exchange rate regimes in emerging Asia. Review of
finance development, 2(2):53-68.
RAPETTI, M. 2011. Macroeconomic Policy Coordination in a Competitive Real
Exchange Rate Strategy for Development, UMASS Amherst Economics Working Papers,
University of Massachusetts Amherst, Department of Economics.
RAZIN, O; COLLINS, S. 1997. Real Exchange Rate Misalignments and Growth.
International Finance 9707001, EconWPA.
RODRIK, D. 2008. The Real Exchange Rate and Economic Growth: Theory and
Evidence, Brooking Papers on economic activity, 39(2):365-439.
WOOLDRIDGE, J. F. 2002. Econometric analysis of cross section and panel data. The
MIT Press, 2002.